Depends on the nature of the questionnaire, who filled it out, and by what methods you used. For example, have you screened for multivariate outliers or for indicators of careless responding (e.g., straightlining, speeding)? Cleaning those things up can increase item correlations. If you’ve done those things, perhaps your items weren’t the best. Were there signs of floor or ceiling effects?

Thank you, Solomon. I think I should check for the outliers

Dear Pallavi,

The EFA process is quite complex. You have only mentioned a few of the aspects. The first stage recommended is to screen for multicollinearity by carrying out a bivariate correlation between all your items, looking for values > 0.8. However, the decision of which member of a pair to remove is qualitative. It is also not quite as simple as looking for pairs as multicollinearity may affect larger clusters of items.

I'm not sure what you are referring to in terms of low correlations of variables in your correlation matrix. Essentially you are trying to establish factors with different items loading on them (typical cut off value is 0.4). The correlation of items within each of these factors should be higher than the correlations between the factors. Even if you use an orthogonal rotation technique, by the time you have chosen items to include in each factor you will end up with non-zero correlations between factors.

Have a look at my guide on EFA: https://www.researchgate.net/publication/304490328_Advice_on_Exploratory_Factor_Analysis

Working Paper Advice on Exploratory Factor Analysis